Organic−Inorganic Heterointerfaces for Ultrasensitive Detection ofUltraviolet LightDali Shao,† Jian Gao,‡ Philippe Chow,‡ Hongtao Sun,§ Guoqing Xin,§ Prachi Sharma,† Jie Lian,§

Nikhil A. Koratkar,*,‡,§ and Shayla Sawyer*,†

†Department of Electrical, Computer, and Systems Engineering, ‡Department of Materials Science and Engineering, and §Departmentof Mechanical, Aerospace, and Nuclear Engineering, Rensselaer Polytechnic Institute, 110 8th Street, Troy, New York 12180, UnitedStates

*S Supporting Information

ABSTRACT: The performance of graphene field-effect transistors islimited by the drastically reduced carrier mobility of graphene on silicondioxide (SiO2) substrates. Here we demonstrate an ultrasensitiveultraviolet (UV) phototransistor featuring an organic self-assembledmonolayer (SAM) sandwiched between an inorganic ZnO quantum dotsdecorated graphene channel and a conventional SiO2/Si substrate.Remarkably, the room-temperature mobility of the chemical-vapor-deposition grown graphene channel on the SAM is an order-of-magnitude higher than on SiO2, thereby drastically reducing electrontransit-time in the channel. The resulting recirculation of electrons (inthe graphene channel) within the lifetime of the photogenerated holes(in the ZnO) increases the photoresponsivity and gain of the transistorto ∼108 A/W and ∼3 × 109, respectively with a UV to visible rejectionratio of ∼103. Our UV photodetector device manufacturing is alsocompatible with current semiconductor processing, and suitable for large volume production.

KEYWORDS: Ultraviolet photodetector, graphene, heterointerface, ZnO quantum dots, gain

Graphene based field-effect transistors (FETs)1−3 arepromising for various applications including photo-

detection, gas sensing, and bioelectronics, which greatly benefitfrom graphene’s notably high mobility, flexibility, and ultrathinnature.4,5 Indeed, extraordinarily high carrier mobilities of up to∼200,000 cm2 V−1 s−1 have been reported in suspendedgraphene samples at 5 K.6 Unfortunately, graphene transistorsfabricated on dielectric substrates (such as SiO2) exhibit a field-effect mobility that is orders of magnitude lower than that ofthe free-standing graphene, which significantly reduces theperformance of such graphene-based FETs.7−11 The reducedmobility of graphene on SiO2 originates from charged impurityscattering,7,8 extrinsic scattering by surface phonons,9 resonantscattering from atomic scale defects,10 and corrugation byresidual adsorbates.11 In order to realize the full potential ofgraphene for next generation electronic and optoelectronicdevices, it is crucial that the harmful effects of the substrate beminimized so that the high mobility achieved in free-standinggraphene can be largely preserved.For device applications, nonpolar substrates have been

suggested to be beneficial,6 yet finding such materials remainsa challenge. For example, scalable methods of deployinghexagonal boron nitride such as chemical vapor depositionresult in significantly wrinkled and defective films12 that do notprovide the same level of benefit as mechanically exfoliatedboron nitride in microelectronics applications. Recent studies

have shown that insertion of an organic self-assembledmonolayer (SAM) between SiO2 and graphene can significantlyimprove the carrier mobility of graphene.13−15 The SAM acts asa spacer that physically separates and electrically decouples thegraphene sheet from the SiO2 surface. However, so far themajority of work has focused on mechanically exfoliatedgraphene that does not provide a viable route to practicaldevices. Further, there is no report of graphene photo-transistors employing SAMs for high-performance photo-sensing applications. In this study, we have utilized a 10-decyltrichlorosilane SAM to support a chemical vapordeposition (CVD) grown monolayer graphene sheet on aSiO2/Si substrate. We report an order of magnitude increase inthe room-temperature carrier mobility of the graphene sheetfrom ∼1,120 cm2 V−1 s−1 without the SAM to ∼10,800 cm2 V−1

s−1 with the SAM. We further deposit ZnO quantum dots(QDs) on the SAM-supported graphene and construct aphototransistor16−18 device for ultraviolet (UV) detection. Theextraordinarily high carrier mobility of the graphene−SAMheterointerface boosts the photoconductive gain of our hybridphototransistor by drastically reducing the electron transit timewithin the FET channel. The phototransistor demonstrates a

Received: January 29, 2015Revised: April 22, 2015Published: May 4, 2015

Letter

pubs.acs.org/NanoLett

© 2015 American Chemical Society 3787 DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

Figure 1. (a) Three-dimensional schematic illustration of the hybrid phototransistor device. (b) Side view of the hybrid phototransistor withillustration of the molecular structure of the organic self-assembled monolayer (SAM) support. (c) Optical microscopy (left) and scanning electronmicroscopy (right) images of the chemical vapor deposition grown monolayer graphene used in this study in contact with the gold electrodes,forming the phototransistor. The scale bar in the figure corresponds to ∼50 μm. (d) Raman spectrum of the monolayer graphene on SAM/SiO2before (blue) and after (red) deposition of ZnO QDs.

Figure 2. (a) Channel resistance of the pristine graphene and the ZnO QDs-graphene measured under dark environment. (b) Channel resistance ofthe ZnO QDs-graphene hybrid phototransistor as a function of back-gate voltage with varying illuminating power (wavelength: 335 nm). Increasingthe illumination leads to a shift of the Dirac point to lower back-gate voltage; this indicates transfer of electrons from the ZnO QDs to the graphenechannel. (c) Channel resistance as a function of illuminating power and back-gate voltage. (d) Shift of Dirac point as a function of the illuminatingpower on the hybrid phototransistor.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

3788

photoresponsivity of ∼108 A/W and a gain of ∼3 × 109 in theUV region with an excellent UV to visible rejection ratio(∼1000). To our knowledge this is the highest photo-responsivity and gain that has been reported in the literaturefor UV sensing and is at least an order of magnitude better thanthe performance of the phototransistor devices without theSAM. The specific detectivity of our device is ∼5.1 × 1013 Jonesat 368 nm at room temperature. For commercial photo-detectors, only photomultiplier tube (PMT) has a higherdetectivity than our device (in the UV region). However, PMThas several obvious disadvantages, including large size, highcost, and the requirement for a high-voltage power supply(typically above 1 kV).In this study, we also provide a detailed physical model to

explain the internal gain mechanism in the heterostructure,which will help to further understand the working mechanismand improve the performance of such graphene-based photo-transistors. In addition to the high sensitivity, the fabricationprocess for our phototransistor device is scalable because weuse large-area CVD graphene and SAM deposition at the waferscale on Si/SiO2 substrates. This process is compatible withsemiconductor processing and could potentially be scaled forproduction at industrially relevant scales. Such ultrasensitiveUV detectors have a wide range of possible applicationsincluding secure space-to-space communications, pollutionmonitoring, water sterilization, flame sensing, and early missileplume detection.17−19

Figure 1a is the three-dimensional (3D) schematicillustration of the hybrid phototransistor. The structure of the10-decyltrichlorosilane SAM used in this study is provided inSupporting Information 1, Figure S1. The channel of thephototransistor, consisting of a monolayer graphene sheetdecorated with ZnO QDs, is placed on top of a SAM/SiO2/Siwafer (see Supporting Information 2 for detailed procedure forsynthesis and deposition of ZnO QDs on SAM supportedgraphene channel). Figure 1b illustrates the charge transportprocess in the hybrid phototransistor when placed under UVillumination, and the molecular structure of the SAM. Figure 1cshows optical microscopy (left) and scanning electronmicroscopy (right) images of the CVD grown monolayergraphene in contact with gold electrodes, forming thephototransistor. The Raman spectra of the graphene transferredto the SAM/SiO2 before and after deposition of ZnO QDs areshown in Figure 1d. For the pristine graphene on SAM, the

absence of a D band peak (∼1350 cm−1) in the Raman spectraconfirms the high quality and defect-free nature of the graphenesample.20 The 2D peak (∼2675 cm−1) is also spectrally sharpand its intensity is about three times greater than that of the Gpeak (∼1583 cm−1), which are characteristics of monolayergraphene.21 After deposition of ZnO QDs, a small D peak(∼1338 cm−1) appeared in the Raman spectrum (red curve),which is due to additional disorder/defects introduced duringthe deposition of QDs.22,23 Ellipsometry analysis gave anaverage thickness of the SAM of ∼1.3 ± 0.1 nm on the SiO2/SiWafer. Transmission electron microscopy (TEM) image andphotoluminescence (PL) spectrum of the ZnO QDs areprovided in Supporting Information 2, Figure S2. The X-raydiffraction (XRD) and optical absorption analysis of the ZnOdecorated graphene can be found in Supporting Information 2,Figure S3.Figure 2a shows the channel resistance of the pristine

graphene on SAM, and the ZnO QDs-graphene on SAM as afunction of backgate voltage (VBG) measured under darkenvironment. The Dirac point VD (i.e., the charge neutralitypoint) of the pristine graphene on SAM was initially at 23.6 Vdue to doping effects from the transfer process and thesubstrate. After deposition of the ZnO QDs on the device, theVD was shifted to about 1.8 V, consistent with a slightly higherwork function of ZnO QDs as compared to graphene, and theelectron carrier mobility (see Methods in SupportingInformation) decreased from ∼10,800 to ∼6,700 cm2 V−1 s−1

due to the additional disorder/defects introduced by theQDs.22,23 The carrier mobility of our hybrid phototransistor isalmost 1 order of magnitude higher than that of a controldevice without the SAM (Supporting Information 3, and FigureS4) for which the mobility before and after deposition of ZnOQDs were measured to be ∼1,120 and ∼630 cm2 V−1 s−1,respectively. Figure 2b shows the channel resistance of ourhybrid phototransistor as a function of VBG under varyingilluminating power. The illumination causes the Dirac point toshift to lower values of VBG, which is due to transfer ofphotogenerated electrons from the ZnO QDs to the graphenechannel (photodoping effect). For a more detailed analysis ofthe channel resistance, we plot the channel resistance as afunction of illuminating power and back-gate voltage, as shownin Figure 2c. From this data, the shift of VD as a function ofilluminating power is extracted. For a low illuminating power,the photodoping induced shift of VD can reach to ∼1.9 V/pW,

Figure 3. (a) Photocurrent of the hybrid phototransistor as a function of drain−source voltage measured under varying illuminating power. Inset:total current of the device (photocurrent + dark current). (b) Photoresponsivity spectrum of our hybrid UV phototransistor (SAM-G-QDs) incomparison with commercial PMT, Si APD and GaN photodetectors.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

3789

demonstrating the high photosensitivity of our device at lowilluminating power intensity. As the illuminating powerincreases, the VD shift per unit optical power decreases,indicating reduced sensitivity for higher illuminating intensity.This is due to the reduced photodoping effect that originatesfrom the saturation of surface trapped holes and reducedaverage lifetime of the holes at high illuminating intensity. Thedetailed analysis of this mechanism will be provided later.The photocurrent as a function of source-to-drain bias (VDS)

is shown in Figure 3a, which reveals that the photocurrentdepends linearly on the bias voltage VDS. The photoresponsivityspectrum of the hybrid phototransistor, defined as thephotocurrent per unit incident optical power, is shown inFigure 3b. For comparison, we also include the photo-responsivity spectra of a Si-avalanche photodiode (APD), aGaN photodetector, and a photomultiplier tube (PMT).Remarkably, the maximum photoresponsivity of our devicewith zero gate bias can reach to ∼1.1 × 108 A/W in the UVregion, which is at least 7 orders of magnitude higher thanthose of Si-APD and GaN photodetectors and is even 10 timeshigher than that of PMTs. Notably our device provides anexcellent UV to visible rejection ratio (∼103) that is 3 orders ofmagnitude higher than that of the PMT. The high photo-

responsivity of our hybrid phototransistor is attributed to thehigh photoconductive gain that originates from the recircula-tion of electrons in the graphene channels within the lifetime ofthe photogenerated holes, which can be understood byreferring to the carrier dynamics in the device, which isillustrated in Figure 4a.It is well established that oxygen molecules adsorb onto

oxide surfaces and capture the free electrons present in n-typemetal oxide semiconductors, [O2(g) + e− → O2

−(ad)], forminga low-conductivity depletion layer near the surface of ZnO, asshown in the top panel of Figure 4a.24−26 For ZnO QDs, thereis usually a high density of hole-trap states on the surface due toits high surface-to-volume ratio,24 and this is in good agreementwith the photoluminescence spectrum measured for the as-prepared ZnO QDs (Supporting Information 2, Figure S2b).When illuminated with photon energy larger than the bandgapof ZnO, electron−hole pairs are generated and soon separatedwith the holes trapped at the surface (due to the potential slopenear the surface due to band bending), leaving behind unpairedelectrons that quickly transfer to the graphene channel and arecollected by the drain by means of the source-drain voltage.The trapped holes recombine with negatively charged oxygenions and neutral oxygen molecules are then desorbed from the

Figure 4. (a) Schematic illustration of the mechanism that contributes to the ultrahigh photoresponsivity of the hybrid phototransistor. (b)Transient response of the device for a excitation at wavelength of 335 nm (power: 138 pW). The red curve is the best-fitting for the data obtainedwith a biexponential decay function. (c) Photoconductive gain versus illuminating power, showing the high sensitivity of the phototransistor. Thetransistor exhibits a gain of ∼3 × 109 for an illuminating power of 1.9 pW and shows a progressive decrease with increasing illuminating power dueto the saturation of the surface states and reduced average lifetime of holes. The inset shows the photoresponsivity of our device as a function of theback-gate voltage, showing that the photoresponsivity of the detector can be directly controlled by the applied back-gate potential and can be tunedfrom ∼2.4 × 107 A/W to 0. The measurement was performed for an excitation at wavelength of 335 nm with a power of ∼138 pW. (d)Photoconductive gain for the ZnO QDs-SAM-G device is compared to a baseline ZnO QDs-G device without the SAM and other published reportson UV photodetectors.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

3790

surface of the ZnO QDs [O2−(ad) + h+→ O2(g)], as shown in

the bottom panel of Figure 4a. The effect of the oxygenmolecule adsorption−desorption process was verified bymeasuring the photoresponse of the device under constantnitrogen flow (Supporting Information 4, Figure S5). Thetransit time of the electrons in the graphene channel is veryshort due to the graphene’s high carrier mobility as well as theshort channel length (20 μm for our device), while the surfaceoxygen molecule adsorption−desorption process usually resultsin a relatively long lifetime of the holes.27 This leads to electronreplenishment from the negative biased source as soon as anelectron reaches the positive biased drain. Accordingly, multipleelectrons will circulate in the graphene channel following asingle electron−hole photogeneration event, leading to a veryhigh photoconductive gain. It is to be noted that oxygen andwater molecules present in air will also physisorb at thegraphene/SiO2 interface and p-dope the graphene channel.However, as shown in Figure 2a the deposition of ZnO QDsresults in a significant shift of the Dirac Point from 23.6 to 1.8V due to the transfer of electrons from the ZnO QDs to thegraphene, which compensates the doping effect of physisorbedoxygen at the graphene/SiO2 interface. Therefore, the oxygenmolecule adsorption/desorption process at the surface of theZnO QDs plays a much more important role in the gainmechanism of this phototransistor.It is important to note that the gain of our hybrid

phototransistor is expected to be a function of the excitationlight intensity due to the effect of the traps states on the surfaceof the ZnO QDs. At low intensity excitation, the holes,separated from photogenerated electron−hole pairs, willoccupy the surface states, thus remaining trapped at the surfaceand will recombine with the negative charged oxygenmolecules. However, as the light intensity increases and moreelectron−hole pairs are generated, the number of availablehole-trap states present at the surface will be diminished. As thetrap states are filled, the energy bands flatten and the extraelectron−hole pairs recombine immediately after they aregenerated in the time scale of several tens of picoseconds.28

Therefore, the extra electron−hole pairs that are generated(after the saturation of surface states) will not participate in thecharge transfer process, and the average lifetime will beshortened. Because the carrier lifetime of the holes is related tothe excitation light intensity as well as the available surfacetraps, it can be expressed as

=+ ( )

T P T( )1

1 PP

nlife

0

Where, Tlife is the carrier lifetime at low excitation intensity, P isthe excitation light intensity on the QDs, P0 is the saturationlight intensity on the QDs in which the all the surface states arefilled, and n is a phenomenological fitting parameter. Thus, thefree carrier concentration (N) in the ZnO QDs can beexpressed as

ην

ην

= =+ ( )

NP

hT P

Ph

T( )1

1 PP

nlife

0

Where η is the internal quantum efficiency of the QDs and hν isthe photon energy of the excitation light. The gain can thus bededucted as

να ν αη= = =

+ ( )G

I h

qPWLqNVW

hqPWL

TT

( )1

1 PP

nph life

transit0

where α is the carrier transfer efficiency from ZnO QDs to thegraphene channel, q is the elementary charge, W is the channelwidth, and V = μVDSL

−1 is the carrier drift velocity in graphenewith the carrier mobility μ and the sample length L. Ttransit =L2μ−1VDS

−1 is the carrier transit time in the graphene channel.For our device, withW = 90 μm, L = 20 μm, VDS = 1.5 V, and μ= 6700 cm2 V−1 s−1 Ttransit is estimated to be ∼0.4 ns. From theabove expression derived for the gain, we conclude that underlow power illumination (P ≪ Po), a very short transit time ofthe electrons in the graphene channel compared with thelifetime of the holes (Ttransit ≪ Tlife) will result in a high gain forthe device.To quantitatively verify the above physical model for the

gain, it is necessary to determine the carrier lifetime of the holesin the device. We measured the transient response of the hybridphototransistor and the result is presented in Figure 4b. Thedecay trace can be approximately fitted by a biexponentialfunction, from which the lifetime of the holes can be extractedas Tlife = 2.3 s. This approach has been widely used and provedto be an efficient way to determine the carriers’ lifetime.22,24,29

It is noteworthy that the lifetime of the holes in the QDs (∼2.3s) is over 9 orders of magnitude greater than the electrontransit time in the FET channel (∼0.4 ns). On the basis of themeasurements of the response time scale, Ttransit and Tlife, weverify that our experimentally observed gain is in goodagreement with the theoretical gain, as shown in Figure 4c.The device demonstrated a maximum gain of 3 × 109 for anilluminating power of 1.9 pW and showed a progressivedecrease with increasing illuminating power due to thesaturation of the surface states and reduced average lifetimeof holes. The solid red curve is the best fitting for the measuredgain with fitting parameters: P0 = 4.5 pW, n = 0.83, and αη =81.7%. The theoretical calculation predicts that a maximumgain as high as ∼4.7 × 109 can be achieved for excitation lightwith power <10 fW. Such a high gain suggests that our hybridphototransistor could be used for single-photon detection. Forcomparison, the gain measured for a control device withoutSAM fabricated using the same processing technique and thegain reported for other high-performance UV photodetectors inrecent references30−32 has been presented in Figure 4d. Indeed,our hybrid UV phototransistors achieved the highest gain ascompared to all other types of UV photodetectors. Themaximum gain-bandwidth product of our hybrid photo-transistor is on the order of 3.3 × 108 Hz (given by theproduct of 4.7 × 109 gain and ∼0.07 Hz of 3 dB bandwidth).Besides gain and photoresponsivity, the specific detectivity

(D*) is also one of the key figure-of-merits for a photodetector,which is expressed as

* =× Δ

DA fNEP

(Jones)

where the NEP is the noise equivalent power in the unit of A/Hz1/2, Δf is bandwidth in hertz, and A is the area of the devicein cm2. The noise level of our phototransistor under darkenvironment was measured to be ∼2 × 10−3 Ω/Hz1/2, whichsets the lower threshold limit of noise-equivalent power at ∼1.4× 10−16 W (see Supporting Information 5 and Figure S6). Thisyields a specific detectivity D* of ∼5.1 × 1013 Jones at 368 nm

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

3791

at room temperature. The detectivity comparisons of ourhybrid phototransistor with commercial photodetectors andvarious other types of photodetectors (see SupportingInformation 6, Figure S7 and Table S1) demonstrate theoutstanding sensitivity of this device. For commercial photo-detectors, only the photomultiplier tube (PMT) has a higherdetectivity than our device (in the UV region). However, aspreviously discussed in the introduction, PMT devices haveseveral limitations that include its large size, high cost,complexity, and the requirement for high-voltage power supply(typically >1 kV). It should be noted that although our hybridUV phototransistor achieves the highest gain reported to date,its detectivity is lower than commercial PMT due to theextremely high conductivity of the SAM supported graphenechannel, which increases the dark current in the device.In addition to the high sensitivity and detectivity, another

advantage of our hybrid phototransistor is that its gain can betuned by varying the back-gate voltage, as shown in the inset ofFigure 4c. A maximum responsivity as high as ∼2.4 × 107 A/Wwas obtained for VBG = 5.6 V for an excitation at wavelength of∼335 nm with a power intensity of ∼138 pW. By tuning theFermi energy close to the Dirac point at VBG = −6.6 V, theresponsivity can be reduced to zero. Such tunability allowscontrol of the on−off state of the photodetector as well asadjustment of the required gain, which is useful for pixelatedimaging applications where the implementation of nanoscalelocal gates enables a locally tunable photoresponse.

■ ASSOCIATED CONTENT*S Supporting InformationDetails of the material synthesis, device fabrication andcharacterization, and detectivity comparisons are given in thissection. The Supporting Information is available free of chargeon the ACS Publications website at DOI: 10.1021/acs.nanolett.5b00380.

■ AUTHOR INFORMATIONCorresponding Authors*E-mail: (N.A.K.) koratn@rpi.edu.*E-mail: (S.S.) sawyes@rpi.edu.Author ContributionsD.S. and J.G. contributed equally in this manuscript.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSN.A.K. acknowledges funding support from the USA NationalScience Foundation (Awards CMMI 1234641 and CMMI1435783). J. L. acknowledges funding support from the USANational Science Foundation (Award: CMMI 1463083). S. S.acknowledges funding support from the Engineering ResearchCenters Program of the National Science Foundation underNSF Cooperative Agreement No. EEC-0812056 and in part byNew York State under NYSTAR contract C130145.

■ REFERENCES(1) Liu, C.-H.; Chang, Y.-C.; Norris, T. B.; Zhong, Z. Nat.Nanotechnol. 2014, 9, 273−278.(2) Rumyantsev, S.; Liu, G.; Shur, M. S.; Potyrailo, R. A.; Balandin, A.A. Nano Lett. 2012, 12, 2294−2298.(3) Mohanty, N.; Berry, V. Nano Lett. 2008, 8, 4469−4476.(4) Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6, 183−191.(5) Schwierz, F. Nat. Nanotechnol. 2010, 5, 487−496.

(6) Bolotin, K. I.; Sikes, K. J.; Hone, J.; Stormer, H. L.; Kim, P. Phys.Rev. Lett. 2008, 101, 096802.(7) Chen, F.; Xia, J.; Tao, N. Nano Lett. 2009, 9, 1621−1625.(8) Chen, J. H.; Jang, C.; Adam, S.; Fuhrer, M. S.; Williams, E. D.;Ishigami, M. Nat. Phys. 2008, 4, 377−381.(9) Chen, J. H.; Jang, C.; Xiao, S. D.; Ishigami, M.; Fuhrer, M. S. Nat.Nanotechnol. 2008, 3, 206−209.(10) Ni, Z. H.; Ponomarenko, L. A.; Nair, R. R.; Yang, R.;Anissimova, S.; Grigorieva, I. V.; Schedin, F.; Blake, P.; Shen, Z. X.;Hill, E. H.; Novoselov, K. S.; Geim, A. K. Nano Lett. 2010, 10, 3868−3872.(11) Ishigami, M.; Chen, J. H.; Cullen, W. G.; Fuhrer, M. S.;Williams, E. D. Nano Lett. 2007, 7, 1643−1648.(12) Gao, J.; Chow, P. K.; Thomas, A. V.; Lu, T.-M.; Borca-Tasciuc,T.; Koratkar, N. Appl. Phys. Lett. 2014, 105, 123108.(13) Lafkioti, M.; Krauss, B.; Lohmann, T.; Zschieschang, U.; Klauk,H.; von Klitzing, K.; Smet, J. H. Nano Lett. 2010, 10, 1149−1153.(14) Liu, Z.; Bol, A. A.; Haensch, W. Nano Lett. 2011, 11, 523−528.(15) Wang, X.; Xu, J.-B.; Wang, C.; Du, J.; Xie, W. Adv. Mater. 2011,23, 2464−2468.(16) Petrone, N.; Dean, C. R.; Meric, I.; van der Zande, A. M.;Huang, P.; Wang, L.; Muller, D.; Shepard, K. L.; Hone, J. Nano Lett.2012, 12, 2751−2756.(17) Liu, S.; Ye, J.; Cao, Y.; Shen, Q.; Liu, Z.; Qi, L.; Guo, X. Small2009, 5, 2371−2376.(18) Zhang, D.; Gan, L.; Cao, Y.; Wang, Q.; Qi, L.; Guo, X. Adv.Mater. 2012, 24, 2715−2720.(19) Li, H.-G.; Wu, G.; Shi, M.-M.; Yang, L.-G.; Chen, H.-Z.; Wang,M. Appl. Phys. Lett. 2008, 93, 153309.(20) Wang, Y.; Ni, Z.; Yu, T.; Shen, Z.; Wang, H.; Wu, Y.; Chen, W.;Wee, A. T. S. J. Phys. Chem. C 2008, 112, 10637.(21) Dresselhaus, M. S.; Jorio, A.; Hofmann, M.; Dresselhaus, G.;Saito, R. Nano Lett. 2010, 10, 751−758.(22) Konstantatos, G.; Badioli, M.; Gaudreau, L.; Osmond, J.;Bernechea, M.; de Arquer, F. P. G.; Gatti, F.; Koppens, F. H. L. Nat.Nanotechnol. 2012, 7, 363−368.(23) Guo, W.; Xu, S.; Wu, Z.; Wang, N.; Loy, M. M. T.; Du, S. Small2013, 9, 3031−3036.(24) Soci, C.; Zhang, A.; Xiang, B.; Dayeh, S. A.; Aplin, D. P. R.;Park, J.; Bao, X. Y.; Lo, Y. H.; Wang, D. Nano Lett. 2007, 7, 1003−1009.(25) Shao, D.; Yu, M.; Sun, H.; Hu, T.; Sawyer, S. Nanoscale 2013, 5,3664−3667.(26) Fan, Z.; Wang, D.; Chang, P.-C.; Tseng, W.-Y.; Lu, J. G. Appl.Phys. Lett. 2004, 85, 5923−5925.(27) Robel, I.; Bunker, B. A.; Kamat, P. V. Adv. Mater. 2005, 17,2458−2463.(28) Fonoberov, V. A.; Balandin, A. A. J. Nanoelectron. Opotelectron.2006, 1, 19−38.(29) Sun, Z.; Liu, Z.; Li, J.; Tai, G.-A.; Lau, S.-P.; Yan, F. Adv. Mater.2012, 24, 5878−5883.(30) Liu, X.; Gu, L.; Zhang, Q.; Wu, J.; Long, Y.; Fan, Z. Nat.Commun. 2014, 5, 4007.(31) Yu, Y.; Jie, J.; Jiang, P.; Wang, L.; Wu, C.; Peng, Q.; Zhang, X.;Wang, Z.; Xie, C.; Wu, D.; Jiang, Y. J. Mater. Chem. 2011, 21, 12632−12638.(32) Guo, F.; Yang, B.; Yuan, Y.; Xiao, Z.; Dong, Q.; Bi, Y.; Huang, J.Nat. Nanotechnol. 2012, 7, 798−802.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.5b00380Nano Lett. 2015, 15, 3787−3792

3792